Re: 923Mbits/s across the ocean
[just discovered in my unsent messages queue from offline composition, probably not timely, but...] Iljitsch van Beijnum wrote: We can't replace path MTU discovery (but hopefully people will start to realize ICMP messages were invented for another reason than job security for firewalls). But what we need is a way for 10/100 Mbps 1500 byte hosts to live with 1000 Mbps 9000 byte hosts on the same subnet. I thought IPv6 neighbor discovery supported this because ND can communicate the MTU between hosts on the same subnet, but unfortunately this is a subnet-wide MTU and not a per-host MTU, which is what we really need. A decade ago, when I designed SIPP Neighbor Discovery, it saved per destination maximum unfragmented datagram size in the route cache, and each I-Am-Here message Heard specified Maximum Receive Unit (MRU) per host. Thus, once upon a time, IPv6 had what you need. Unfortunately, the IPv6 group stripped out such innovative features. I stopped paying attention after the new editor stated something like it worked for ethernet, we really don't need any more than that. Well, we used IPv4 from '83, and designed SIPP (cum IPv6) in '93. IPv6 is a failure -- maybe it's time for this decade's design? Or maybe even some of the features some of us thought we needed a decade ago? -- William Allen Simpson Key fingerprint = 17 40 5E 67 15 6F 31 26 DD 0D B9 9B 6A 15 2C 32
Re: 923Mbits/s across the ocean
On Mon, 10 Mar 2003, Richard A Steenbergen wrote: On the receive size, the socket buffers must be large enough to accommodate all the data received between application read()'s, That's not true. It's perfectly acceptable for TCP to stall when the receiving application fails to read the data fast enough. Ok, I think I was unclear. You don't NEED to have buffers large enough to accommodate all that data received between application read()'s, unless you are trying to achieve maximum performance. I thought that was the general framework we were all working under. :) You got me there. :-) It seemed that you were talking about more general requirements at this point, though with the upper and lower limits for kernel buffer space and all. Hm, I don't see this happening to a usable degree as TCP has no concept of records. You really want to use fixed size chunks of information here rather than pretending everything's a stream. We're talking optimizations for high performance transfers... It can't always be a stream. Right. But TCP is a stream protocol. This has many advantages, nearly all of which are irrelevant for high volume high bandwidth bulk data transfer. I can imagine a system that only works in one direction and where the data is split into fixed size records (which would ideally fit into a single packet) where each record is acknowledged independently (but certainly not for each individual packet). I would also want to take advantage of traffic classification mechanisms: first the data is flooded at the maximum speed at the lowest possible traffic class. Everything that doesn't make it to the other end is then resent slower with a higher traffic class. If the network supports priority queuing then this would effectively sponge up all free bandwidth without impacting regular interactive traffic. If after a few retries some data still didn't make it: simply skip this for now (but keep a record of the missing bits) and keep going. Many applications can live with some lost data and for others it's probably more efficient to keep running at high speed and repair the gaps afterwards. IMHO the 1500 byte MTU of ethernet will still continue to prevent good end to end performance like this for a long time to come. But alas, I digress... Don't we all? I'm afraid you're right. Anyone up for modifying IPv6 ND to support a per-neighbor MTU? This should make backward-compatible adoption of jumboframes a possibility. (Maybe retrofit ND into v4 while we're at it.) Not necessarily sure thats the right thing to do, but SOMETHIG has got to be better than what passes for path mtu discovery now. :) We can't replace path MTU discovery (but hopefully people will start to realize ICMP messages were invented for another reason than job security for firewalls). But what we need is a way for 10/100 Mbps 1500 byte hosts to live with 1000 Mbps 9000 byte hosts on the same subnet. I thought IPv6 neighbor discovery supported this because ND can communicate the MTU between hosts on the same subnet, but unfortunately this is a subnet-wide MTU and not a per-host MTU, which is what we really need. Iljitsch
Re: 923Mbits/s across the ocean
Thus spake Iljitsch van Beijnum [EMAIL PROTECTED] This is the part about TCP that I've never understood: why does it send large numbers of packets back-to-back? This is almost never a good idea. Because until you congest the network to the point of dropping packets, a host has no idea how much bw is actually available. Exponential rate growith finds this value very quickly. Hm, I don't see this happening to a usable degree as TCP has no concept of records. You really want to use fixed size chunks of information here rather than pretending everything's a stream. A record-oriented, reliable transport would make many protocols much easier to implement. Too bad SCTP hasn't seen wider use. S Stephen Sprunk God does not play dice. --Albert Einstein CCIE #3723 God is an inveterate gambler, and He throws the K5SSSdice at every possible opportunity. --Stephen Hawking
Re: 923Mbits/s across the ocean
On Sat, 2003-03-08 at 15:58, [EMAIL PROTECTED] wrote: That's the argument that pentagon used to justify buying $40 lightbulbs. Does not work, sorry. That is not the argument used to justify buying 40 lightbulbs. They do not actually purchase 40 lightbulbs, the prices that you see in rag magazine reports has to do with how the budgets are handled. If you can budget a multi-billion dollar organization and put in reasonable price and performance controls there are many schools that would hire you after you revolutionized public administration and the DoD... -- Douglas F. Calvert [EMAIL PROTECTED]
Re: 923Mbits/s across the ocean
On Sun, 9 Mar 2003, Richard A Steenbergen wrote: On the send size, the application transmitting is guaranteed to utilize the buffers immediately (ever seen a huge jump in speed at the beginning of a transfer, this is the local buffer being filled, and the application has no way to know if this data is going out to the wire, or just to the kernel). Then the network must drain the packets onto the wire, sometimes very slowly (think about a dialup user downloading from your GigE server). Actually this is often way too fast as the congestion window doubles with each ACK. This means that with a large buffer = large window and a bottleneck somewhere along the way, you are almost guaranteed to have some serious congestion in the early stages of the session and lower levels of congestion periodially later on whenever TCP tries to figure out how large the congestion window can get without losing packets. This is the part about TCP that I've never understood: why does it send large numbers of packets back-to-back? This is almost never a good idea. On the receive size, the socket buffers must be large enough to accommodate all the data received between application read()'s, That's not true. It's perfectly acceptable for TCP to stall when the receiving application fails to read the data fast enough. (TCP then simply announces a window of 0 to the other side so the communication effectively stops until the application reads some data and a 0 window is announced.) If not, the kernel would be required to buffer unlimited amounts of data in the event an application fails to read it from the buffer for some time (which is a very common situation). locally. Jumbo frames help too, but their real benefit is not the simplistic hey look theres 1/3rd the number of frames/sec view that many people see. The good stuff comes from techniques like page flipping, where the NIC DMA's data into a memory page which can be flipped through the system straight to the application, without copying it throughout. Some day TCP may just be implemented on the NIC itself, with ALL work offloaded, and the system doing nothing but receiving nice page-sized chunks of data at high rates of speed. Hm, I don't see this happening to a usable degree as TCP has no concept of records. You really want to use fixed size chunks of information here rather than pretending everything's a stream. IMHO the 1500 byte MTU of ethernet will still continue to prevent good end to end performance like this for a long time to come. But alas, I digress... Don't we all? I'm afraid you're right. Anyone up for modifying IPv6 ND to support a per-neighbor MTU? This should make backward-compatible adoption of jumboframes a possibility. (Maybe retrofit ND into v4 while we're at it.) Iljitsch van Beijnum
Re: 923Mbits/s across the ocean
On Tue, Mar 11, 2003 at 12:41:15AM +0100, Iljitsch van Beijnum wrote: On the receive size, the socket buffers must be large enough to accommodate all the data received between application read()'s, That's not true. It's perfectly acceptable for TCP to stall when the receiving application fails to read the data fast enough. (TCP then simply announces a window of 0 to the other side so the communication effectively stops until the application reads some data and a 0 window is announced.) If not, the kernel would be required to buffer unlimited amounts of data in the event an application fails to read it from the buffer for some time (which is a very common situation). Ok, I think I was unclear. You don't NEED to have buffers large enough to accommodate all that data received between application read()'s, unless you are trying to achieve maximum performance. I thought that was the general framework we were all working under. :) locally. Jumbo frames help too, but their real benefit is not the simplistic hey look theres 1/3rd the number of frames/sec view that many people see. The good stuff comes from techniques like page flipping, where the NIC DMA's data into a memory page which can be flipped through the system straight to the application, without copying it throughout. Some day TCP may just be implemented on the NIC itself, with ALL work offloaded, and the system doing nothing but receiving nice page-sized chunks of data at high rates of speed. Hm, I don't see this happening to a usable degree as TCP has no concept of records. You really want to use fixed size chunks of information here rather than pretending everything's a stream. We're talking optimizations for high performance transfers... It can't always be a stream. IMHO the 1500 byte MTU of ethernet will still continue to prevent good end to end performance like this for a long time to come. But alas, I digress... Don't we all? I'm afraid you're right. Anyone up for modifying IPv6 ND to support a per-neighbor MTU? This should make backward-compatible adoption of jumboframes a possibility. (Maybe retrofit ND into v4 while we're at it.) Not necessarily sure thats the right thing to do, but SOMETHIG has got to be better than what passes for path mtu discovery now. :) -- Richard A Steenbergen [EMAIL PROTECTED] http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)
Re: 923Mbits/s across the ocean
On Sat, 8 Mar 2003, Joe St Sauver wrote: you will see that for bulk TCP flows, the median throughput is still only 2.3Mbps. 95th%-ile is only ~9Mbps. That's really not all that great, throughput wise, IMHO. Strange. Why is that? RFC 1323 is widely implemented, although not widely enabled (and for good reason: the timestamp option kills header compression so it's bad for lower-bandwidth connections). My guess is that the OS can't afford to throw around MB+ size buffers for every TCP session so the default buffers (which limit the windows that can be used) are relatively small and application programmers don't override the default.
RE: 923Mbits/s across the ocean
Also as the OS's are shipped they come with small default maximum window sizes (I think Linux is typically 64KB and Solaris is 8K), and so one has to get the sysadmin with root privs to change this. -Original Message- From: Iljitsch van Beijnum [mailto:[EMAIL PROTECTED] Sent: Sunday, March 09, 2003 5:25 AM To: Joe St Sauver Cc: [EMAIL PROTECTED] Subject: Re: 923Mbits/s across the ocean On Sat, 8 Mar 2003, Joe St Sauver wrote: you will see that for bulk TCP flows, the median throughput is still only 2.3Mbps. 95th%-ile is only ~9Mbps. That's really not all that great, throughput wise, IMHO. Strange. Why is that? RFC 1323 is widely implemented, although not widely enabled (and for good reason: the timestamp option kills header compression so it's bad for lower-bandwidth connections). My guess is that the OS can't afford to throw around MB+ size buffers for every TCP session so the default buffers (which limit the windows that can be used) are relatively small and application programmers don't override the default.
Re: 923Mbits/s across the ocean
On Sun, Mar 09, 2003 at 02:25:25PM +0100, Iljitsch van Beijnum quacked: On Sat, 8 Mar 2003, Joe St Sauver wrote: you will see that for bulk TCP flows, the median throughput is still only 2.3Mbps. 95th%-ile is only ~9Mbps. That's really not all that great, throughput wise, IMHO. Strange. Why is that? RFC 1323 is widely implemented, although not widely enabled (and for good reason: the timestamp option kills header compression so it's bad for lower-bandwidth connections). My guess is that the OS can't afford to throw around MB+ size buffers for every TCP session so the default buffers (which limit the windows that can be used) are relatively small and application programmers don't override the default. Which makes it doubly a shame that the adaptive buffer tuning tricks haven't made it into production systems yet. It was a beautiful, simple idea that worked very well for adapting to long fat networks: http://www.acm.org/sigcomm/sigcomm98/tp/abs_26.html -dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: 923Mbits/s across the ocean
On Sun, Mar 09, 2003 at 08:29:16AM -0800, Cottrell, Les wrote: Strange. Why is that? RFC 1323 is widely implemented, although not widely enabled (and for good reason: the timestamp option kills header compression so it's bad for lower-bandwidth connections). My guess is that the OS can't afford to throw around MB+ size buffers for every TCP session so the default buffers (which limit the windows that can be used) are relatively small and application programmers don't override the default. Also as the OS's are shipped they come with small default maximum window sizes (I think Linux is typically 64KB and Solaris is 8K), and so one has to get the sysadmin with root privs to change this. This is related to how the kernel/user model works in relation to TCP. TCP itself happens in the kernel, but the data comes from userland through the socket interface, so there is a socket buffer in the kernel which holds data coming from and going to the application. TCP cannot release data from it's buffer until it has been acknowledged by the other side, incase it needs to retransmit. This means TCP performance is limited by the smaller of either the congestion window (determined by measuring conditions along the path), or the send/recv window (determined by local system resources). However, you can't just blindly turn up your socket buffers to large values and expect good results. On the send size, the application transmitting is guaranteed to utilize the buffers immediately (ever seen a huge jump in speed at the beginning of a transfer, this is the local buffer being filled, and the application has no way to know if this data is going out to the wire, or just to the kernel). Then the network must drain the packets onto the wire, sometimes very slowly (think about a dialup user downloading from your GigE server). Setting the socket buffers too high can potentially result in an incredible waste of resources, and can severely limit the number of simultaneous connections your server can support. This is precisely why OS's cannot ship with huge default values, because what may be appropriate for your one-user GigE connected box might not be appropriate for someone else's 100BASE-TX web server (and guess which setup has more users :P). On the receive size, the socket buffers must be large enough to accommodate all the data received between application read()'s, as well as making sure they have enough available space to hold future data in the event of a gap due to loss and the need for retransmission. However, if the application fails to read() the data from the socket buffer, it will sit there forever. Large socket buffers also opens the server up to malicious attack causing non-swapable kernel memory to consume all available resources, either locally (by someone dumping data over lots of connections, or running an application which intentionally fails to read data from the socket buffer), or remotely (think someone opening a bunch of rate limited connections from your high speed server). It can even be unintentional, but just as bad (think a million confused dialup users accidentally clicking on your high speed video stream). Some of this can be worked around by implementing what is called auto-tuning socket buffers. In this case, the kernel would limit the amount of data allowed into the buffer, by looking at the tcp session's observed congestion window. This allows you to define large send buffers without applications connected to slow receivers sucking up unnecessary resourced. PSC has had example implementations for quite a while, and recently FreeBSD even added this (sysctl net.inet.tcp.inflight_enable=1 as of 4.7). Unfortunately, there isn't much you can do to prevent malicious receive-side buffer attacks, short of limiting the overall max buffer (FreeBSD implements this as an rlimit sbsize). Of course, you need a few other things before you can start getting into end to end gigabit speeds. If you're transfering a file, you probably don't want to be reading it from disk via the kernel just to send it back to the kernel again for transmission, so various things like sendfile() and zero copy implementations help get you the performance you need locally. Jumbo frames help too, but their real benefit is not the simplistic hey look theres 1/3rd the number of frames/sec view that many people see. The good stuff comes from techniques like page flipping, where the NIC DMA's data into a memory page which can be flipped through the system straight to the application, without copying it throughout. Some day TCP may just be implemented on the NIC itself, with ALL work offloaded, and the system doing nothing but receiving nice page-sized chunks of data at high rates of speed. IMHO the 1500 byte MTU of ethernet will still continue to prevent good end to end performance like this for a long time to come. But alas, I digress... -- Richard A Steenbergen [EMAIL
Re: 923Mbits/s across the ocean
I am not normally on this list but someone kindly gave me copies of some of the email concerning the Internet2 Land Speed record. So I have joined the list. As one of the PIs of the record, I thought it might be useful to comment on a few interesting items I have seen, and no I am not trying to flame anybody: Give em a million dollars, plus fiber from here to anywhere and let me muck with the TCP algorith, and I can move a GigE worth of traffic too - Dave You are modest in your budgetary request. Just the Cisco router (GSR 12406) we had on free loan listed at close to a million dollars, and the OC192 links just from Sunnyvale to Chicago would have cost what was left of the million/per month. We used a stock TCP (Linux kernel TCP). We did however, use jumbo frames (9000Byte MTUs). In response Richard A Steenbergen we are not now living in a tropical foreign country, with lots and lots of drugs and women but then the weather in California is great today. What am I missing here, theres OC48=2.4Gb, OC192=10Gb ... We were running host to host (end-to-end) with a single stream with common off the shelf equipment, there are not too many (I think none) 1GE host NICs available today that are in production (e.g. without signing a non-disclosure agreement). Production commercial networks ... Blow away these speeds on a regular basis. See the above remark about end-to-end application to application, single stream. So, you turn down/off all the parts of TCP that allow you to share bandwidth ... We did not mess with the TCP stack, it was stock off the shelf. ... Mention that Internet speed records are measured in terabit-meters/sec. You are correct, this is important, but reporters want a sound bite and typically only focus on one thing at a time. I will make sure next time I talk to a reporter to emphasize this. Maybe we can get some mileage out of Petabmps (Peta bit metres per second) sounds What kind of production environment needs a single TCP stream of data at 1Gbits/s over a 150ms latency link? Today High Energy Particle Physics needs hundreds of Megabits/s between California and Europe (Lyon, Padova and Oxford) to deliver data on a timely basis form an experiment site at SLAC to regional computer sites in Europe. Today on production acadmeic networks (with sustainable rates of 100 to a few hundred Mbits/s) it takes about a day to transmit just over a Tbyte of data which just about keeps up with the data rates. The data generation rates are doubling / year so within 1-3 years we will be needing speeds like in the record on a production basis. We needed to ensure we can achieve the needed rates, and whether we can do it with off the shelf hardware, how the hosts and OS' need configuring, how to tune the TCP stack or how newer stacks perform, what are the requirements for jumbo frames etc. Besides High Energy Physics other sciences are beginning to grapple with how to repliacte large databases across the globe, such sciences include radio-astronmoy, human genome, global weather, seismic ... The spud gun is interesting, given the distances, probably a 747 freightliner packed with DST tapes or disks is a better idea. Assuming we fill the 747 with say 50 Gbps tapes (disks would probably be better), then if it takes 10 hours to fly from San Francisco (BTW Sunnyvale is near San Francisco not near LA as one person talking about retiring to better weather might lead one to believe) the bandwidth is about 2-4 Tbits/s. However, this ignores the reality of labelling, writing the tapes, removing from silo robot, pocaking, getting to airport, loading, unloading, getting through customs etc. In reality the latency is really closer to 2 weeks. Even worse if there is an error (heads not aligned etc.) then the the retry latency is long and the effort involved considerable. Also the network solution lends itself much better to automation, in our case we saved a couple of full time equivalent people at the sending site to distribute the data on a regular basis to our collaborator sites in France, UK and Italy. The remarks about window size and buffer are interesting also. It is true large windows are needed. To approach 1Gbits/s we require 40MByte windows. If this is going to be a problem, then we need to raise question like this soon and figure out how to address (add more memory, use other protocols etc.). In practice to approcah 2.5Gbits/s requires 120MByte windows. I am quite happy to concede that this does not need to be about some jocks beating a record. I do think it is important to catch the public's attention to why high speeds are important, that they are achievable today application to application (it would also be useful to estimate when such speeds are available to universities, large companies, small companies, the home etc.), and for techies it is important to start to understand the challenges the high speeds raise, e.g. cpu and router memories,
Re: 923Mbits/s across the ocean
LC Date: Sat, 08 Mar 2003 10:04:20 -0800 LC From: Cottrell, Les LC The remarks about window size and buffer are interesting LC also. It is true large windows are needed. To approach LC 1Gbits/s we require 40MByte windows. If this is going to be LC a problem, then we need to raise question like this soon and LC figure out how to address (add more memory, use other LC protocols etc.). In practice to approcah 2.5Gbits/s requires LC 120MByte windows. Yup. About 2x to 2.5x the bandwidth*delay product. I'm still curious about insane SACK or maybe NACK. Spray TCP packets hoping they arrive (good odds), and wait to hear what made or didn't make it. Let the receiving end have the large buffers... sending machines generally must handle a greater number of sessions. ECN also would be a nice way of telling a sender to back off, [hopefully] proactively avoiding packet loss. It certainly seems a shame to require big sending buffers and slow down entire streams just in case a small bit gets lost. Eddy -- Brotsman Dreger, Inc. - EverQuick Internet Division Bandwidth, consulting, e-commerce, hosting, and network building Phone: +1 (785) 865-5885 Lawrence and [inter]national Phone: +1 (316) 794-8922 Wichita ~ Date: Mon, 21 May 2001 11:23:58 + (GMT) From: A Trap [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Please ignore this portion of my mail signature. These last few lines are a trap for address-harvesting spambots. Do NOT send mail to [EMAIL PROTECTED], or you are likely to be blocked.
Re: 923Mbits/s across the ocean
You are modest in your budgetary request. Just the Cisco router (GSR 12406) we had on free loan listed at close to a million dollars, and the OC192 links just from Sunnyvale to Chicago would have cost what was left of the million/per month. No, your budget folks have no clue, which they clearly demonstrate. Anyone here who buys Cisco at the list prices works for companies that for some reason want to waste money. We pay about 10c on a dollar. Anyone leasing OC-192 at that price as opposite to lighting it up is smoking. What am I missing here, theres OC48=2.4Gb, OC192=10Gb ... We were running host to host (end-to-end) with a single stream with common off the shelf equipment, there are not too many (I think none) 1GE host NICs available today that are in production (e.g. without signing a non-disclosure agreement). Again, if this is all available today, what is so new that you guys have done, apart from blowing tons of money? The remarks about window size and buffer are interesting also. It is true large windows are needed. To approach 1Gbits/s we require 40MByte windows. If this is going to be a problem, then we need to raise question like this soon and figure out how to address (add more memory, use other protocols etc.). In practice to approcah 2.5Gbits/s requires 120MByte windows. I am quite happy to concede that this does not need to be about some jocks beating a record. I do think it is important to catch the public's attention to why high speeds are important, that they are achievable today application to application (it would also be useful to estimate when such speeds are available to universities, large companies, small companies, the home etc.), and for techies it is important to start to understand the challenges the high speeds raise, e.g. cpu and router memories, bugs in TCP, OS, application etc., new TCP stacks, new (possibly UDP based) protocols such as tsunami, need for 64 bit counters in monitoring, effects of the NIC card, jumbo requirements etc., and what is needed to address them. Also to try and put it in meaningful terms (such as 2 full length DVD movies in a minute, that could also increase the cease and desist legal messages shipped ;-)) is important. High speeds are not important. High speeds at a *reasonable* cost are important. What you are describing is a high speed at an *unreasonable* cost. Alex
Re: 923Mbits/s across the ocean
On Sat, Mar 08, 2003 at 03:29:56PM -0500, [EMAIL PROTECTED] quacked: High speeds are not important. High speeds at a *reasonable* cost are important. What you are describing is a high speed at an *unreasonable* cost. To paraphrase many a california sufer, dude, chill out. The bleeding edge of performance in computers and networks is always stupidly expensive. But once you've achieved it, the things you did to get there start to percolate back into the consumer stream, and within a few years, the previous bleeding edge is available in the current O(cheap) hardware. A cisco 7000 used to provide the latest and greatest performance in its day, for a rather considerable cost. Today, you can get a box from Juniper for the same price you paid for your 7000 that provides a few orders of magnitude more performance. But to get there, you have to be willing to see what happens when you push the envelope. That's the point of the LSR, and a lot of other research efforts. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: 923Mbits/s across the ocean
To paraphrase many a california sufer, dude, chill out. When the none of my taxes goes to the silly projects, I will chill out. It had been stated by the people that participated in this research that (a) they bought hardware at the prices to help Cisco to make its quarters (b) they have spent millions of dollars for OC-192 links when they did not need them. (c) they did not come up with anything new apart from a proof that they achieved that speed. The bleeding edge of performance in computers and networks is always stupidly expensive. But once you've achieved it, the things you did to get there start to percolate back into the consumer stream, and within a few years, the previous bleeding edge is available in the current O(cheap) hardware. That is all great if they *actually* *developed* something. However, they did not. They bought off the shelf products for list prices plugged them in, ran slightly tweaked kernels, helped Qwest/Globalcrossing etc prop its quarters and announced we did it. A cisco 7000 used to provide the latest and greatest performance in its day, for a rather considerable cost. Today, you can get a box from Juniper for the same price you paid for your 7000 that provides a few orders of magnitude more performance. But to get there, you have to be willing to see what happens when you push the envelope. That's the point of the LSR, and a lot of other research efforts. That's the argument that pentagon used to justify buying $40 lightbulbs. Does not work, sorry. Alex
RE: 923Mbits/s across the ocean
With the glossing over of details that goes with press releases there appears to be a misunderstanding here. I never said we paid list prices. I am well aware that one can get large discounts from vendors. However, I think it is important to quote a well known price (in this case list), which people can relate to how well they think they can negotiate (otherwise it just becomes a bragging point of who can get the largest discount), and gets away from the point of giving people an idea of what it might cost. In our case we got 100% (free) discounts from Level(3) and Cisco for the Sunnyvale to Chicago link and the GSR. The link from StarLight to Amsterdam was put in place for a European funded demonstration (since turned into a production link), the equipment was mainly funded by another European research project. At the same time, getting it for free has its costs, one has much less leverage with the vendors as to delivery (and retrieval) dates, reliability etc. as well as the headaches of getting everything (PCs, loaned NIC cards, Routers, links) to come together, to keep the vendors interest, extend the loan etc. High speed at reasonable costs are the end-goal. However, it is important to be able to plan for when one will need such links, to know what one will be able to achieve, and for regular users to be ready to use them when the commonly available. This takes some effort up front to achieve and demonstrate. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Sent: Saturday, March 08, 2003 12:30 PM To: Cottrell, Les Cc: '[EMAIL PROTECTED]' Subject: Re: 923Mbits/s across the ocean You are modest in your budgetary request. Just the Cisco router (GSR 12406) we had on free loan listed at close to a million dollars, and the OC192 links just from Sunnyvale to Chicago would have cost what was left of the million/per month. No, your budget folks have no clue, which they clearly demonstrate. Anyone here who buys Cisco at the list prices works for companies that for some reason want to waste money. We pay about 10c on a dollar. Anyone leasing OC-192 at that price as opposite to lighting it up is smoking. What am I missing here, theres OC48=2.4Gb, OC192=10Gb ... We were running host to host (end-to-end) with a single stream with common off the shelf equipment, there are not too many (I think none) 1GE host NICs available today that are in production (e.g. without signing a non-disclosure agreement). Again, if this is all available today, what is so new that you guys have done, apart from blowing tons of money? The remarks about window size and buffer are interesting also. It is true large windows are needed. To approach 1Gbits/s we require 40MByte windows. If this is going to be a problem, then we need to raise question like this soon and figure out how to address (add more memory, use other protocols etc.). In practice to approcah 2.5Gbits/s requires 120MByte windows. I am quite happy to concede that this does not need to be about some jocks beating a record. I do think it is important to catch the public's attention to why high speeds are important, that they are achievable today application to application (it would also be useful to estimate when such speeds are available to universities, large companies, small companies, the home etc.), and for techies it is important to start to understand the challenges the high speeds raise, e.g. cpu and router memories, bugs in TCP, OS, application etc., new TCP stacks, new (possibly UDP based) protocols such as tsunami, need for 64 bit counters in monitoring, effects of the NIC card, jumbo requirements etc., and what is needed to address them. Also to try and put it in meaningful terms (such as 2 full length DVD movies in a minute, that could also increase the cease and desist legal messages shipped ;-)) is important. High speeds are not important. High speeds at a *reasonable* cost are important. What you are describing is a high speed at an *unreasonable* cost. Alex
RE: 923Mbits/s across the ocean
With the glossing over of details that goes with press releases there appears to be a misunderstanding here. I never said we paid list prices. I am well aware that one can get large discounts from vendors. However, I think it is important to quote a well known price (in this case list), which people can relate to how well they think they can negotiate (otherwise it just becomes a bragging point of who can get the largest discount), and gets away from the point of giving people an idea of what it might cost. In our case we got 100% (free) discounts from Level(3) and Cisco for the Sunnyvale to Chicago link and the GSR. Ok, after such explanation, I am more than willing to accept that it could be a good use of the money, including the money that was paid to people to sit and tweak parameters of gear, kernels, NIC cards to achieve imporovements in speed (since no one in production world can justify having people on the clock doing just that to document the smallest possible improvements). High speed at reasonable costs are the end-goal. However, it is important to be able to plan for when one will need such links, to know what one will be able to achieve, and for regular users to be ready to use them when the commonly available. This takes some effort up front to achieve and demonstrate. True, however as it was mentioned before, why not do the same type of testing in a lab environment between a couple of boxes having the TCP stack insert appropriate delays? When in 1995 we were getting simplex IP links over satellites up that is how we did the testing before bringing them up on the birds. Alex
RE: 923Mbits/s across the ocean
LC Date: Sat, 08 Mar 2003 13:13:53 -0800 LC From: Cottrell, Les LC The link from StarLight to Amsterdam was put in place for a man 4 dummynet LC High speed at reasonable costs are the end-goal. However, it LC is important to be able to plan for when one will need such LC links, to know what one will be able to achieve, and for LC regular users to be ready to use them when the commonly LC available. This takes some effort up front to achieve and LC demonstrate. The thing is we already know that large buffers help greatly. Seeing how fast one can push a box with big buffers might be cool, but is it accomplishing anything? As you demonstrated, anyone who needs that speed here and now can get a private line and use a stock *ix install. Done/done. How about other models? Limited server buffers (it's nice to handle more than 25 simultaneous streams), random-bandwidth clients, congestion, jitter... how were those treated? Have these been explored? If there's going to be research, let's see some TCP stack tuning and the results. Investigating other protocols would be nice; perhaps the scope of the contest should be changed. The level of research in unleashing bone-stock equipment is more appropriate for an undergrad paper than a news release. Eddy -- Brotsman Dreger, Inc. - EverQuick Internet Division Bandwidth, consulting, e-commerce, hosting, and network building Phone: +1 (785) 865-5885 Lawrence and [inter]national Phone: +1 (316) 794-8922 Wichita ~ Date: Mon, 21 May 2001 11:23:58 + (GMT) From: A Trap [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Please ignore this portion of my mail signature. These last few lines are a trap for address-harvesting spambots. Do NOT send mail to [EMAIL PROTECTED], or you are likely to be blocked.
Re: 923Mbits/s across the ocean
On Sat, 8 Mar 2003, Cottrell, Les wrote: We used a stock TCP (Linux kernel TCP). We did however, use jumbo frames (9000Byte MTUs). What kind of difference did you see as opposed to standard 1500 byte packets? I did some testing once and things actually ran slightly faster with 1500 byte packets, completely contrary to my expectations... (This was UDP and just 0.003 km rather than 10,000, though.) The remarks about window size and buffer are interesting also. It is true large windows are needed. To approach 1Gbits/s we require 40MByte windows. If this is going to be a problem, then we need to raise question like this soon and figure out how to address (add more memory, use other protocols etc.). In practice to approcah 2.5Gbits/s requires 120MByte windows. So how much packet loss did you see? Even with a few packets in a million lost this would bring your transfer way down and/or you'd need even bigger windows. However, bigger windows mean more congestion. When two of those boxes start pushing traffic at 1 Gbps with a 40 MB window, you'll see 20 MB worth of lost packets due to congestion in a single RTT. A test where the high-bandwidth session or several high-bandwidth sessions have to live side by side with other traffic would be very interesting. If this works well it opens up possibilities of doing this type of application over real networks rather than (virtual) point-to-point links where congestion management isn't an issue.
RE: 923Mbits/s across the ocean
High speed at reasonable costs are the end-goal. However, it is important to be able to plan for when one will need such links, to know what one will be able to achieve, and for regular users to be ready to use them when the commonly available. This takes some effort up front to achieve and demonstrate. True, however as it was mentioned before, why not do the same type of testing in a lab environment between a couple of boxes having the TCP stack insert appropriate delays? When in 1995 we were getting simplex IP links over satellites up that is how we did the testing before bringing them up on the birds. Following up on and driven by the work leading up to and following the Land Speed Record, some of the Caltech people collaborating on this record together with collaborators from SLAC and elsewhere, are proposing a WAN in the Lab that can be used for just such testing. This saves on leasing fibers but there are still considerable expenses to run at 10Gbit/s rates (cpus, NICs, optical multiplexing equipment etc.). It is also a much more controlled environment that simplifies things. On the other hand it misses out on the real world experience, and so eventually has to be tested first on real world lightly used testbeds, and then on advanced research networks and finally on production networks, to understand how issues such as fairness, congestion avoidance, robustness to poor implemementations or configurations etc. really work.
RE: 923Mbits/s across the ocean
The jumbo frames effectively increase the congestion avoidance additive increase of the congestion avoidance phase of TCP by a factor of 6. Thus after a congestion event, that reduces the window by a factor of 2, one can recover 6 times as fast. This is very important on large RTT fast links where the recovery rate(for TCP/Reno) goes as the MTU/RTT^2. This can be seen in some of the graphs at: http://www-iepm.slac.stanford.edu/monitoring/bulk/fast/stacks.png or more fully at: http://www-iepm.slac.stanford.edu/monitoring/bulk/fast/ We saw little congestion related packet loss on the testbed. With big windows SACK becomes increasingly important so one does not have recover a large fraction of the window for a single packet. Once one gets onto networks where one is really sharing the bandwidth with others performance drops off rapidly (see for example the measuremsnts at http://www-iepm.slac.stanford.edu/monitoring/bulk/fast/#Measurements%20from%20Sunnyvale%20to%20Amsterdam and compare them with those at http://www-iepm.slac.stanford.edu/monitoring/bulk/fast/#TCP%20Stack%20Comparisons%20with%20Single%20Streams One of the next things we want to look at next is how the various new TCP stacks work on production Academic Research Networks (e.g. from Internet2, ESnet, GEANT, ...) with lots of other competing traffic. -Original Message- From: Iljitsch van Beijnum [mailto:[EMAIL PROTECTED] Sent: Saturday, March 08, 2003 1:49 PM To: Cottrell, Les Cc: '[EMAIL PROTECTED]' Subject: Re: 923Mbits/s across the ocean On Sat, 8 Mar 2003, Cottrell, Les wrote: We used a stock TCP (Linux kernel TCP). We did however, use jumbo frames (9000Byte MTUs). What kind of difference did you see as opposed to standard 1500 byte packets? I did some testing once and things actually ran slightly faster with 1500 byte packets, completely contrary to my expectations... (This was UDP and just 0.003 km rather than 10,000, though.) The remarks about window size and buffer are interesting also. It is true large windows are needed. To approach 1Gbits/s we require 40MByte windows. If this is going to be a problem, then we need to raise question like this soon and figure out how to address (add more memory, use other protocols etc.). In practice to approcah 2.5Gbits/s requires 120MByte windows. So how much packet loss did you see? Even with a few packets in a million lost this would bring your transfer way down and/or you'd need even bigger windows. However, bigger windows mean more congestion. When two of those boxes start pushing traffic at 1 Gbps with a 40 MB window, you'll see 20 MB worth of lost packets due to congestion in a single RTT. A test where the high-bandwidth session or several high-bandwidth sessions have to live side by side with other traffic would be very interesting. If this works well it opens up possibilities of doing this type of application over real networks rather than (virtual) point-to-point links where congestion management isn't an issue.